Brief summary of the bug. What is it ? Where is it ?

While investigating an incident we noticed an unusal amount of requests to the process overview index. This appears to be caused by incorrectly built bulk request that due to its placement in a loop continues to grow in size, baiscally resulting in a fibonacci sequence of bulk requests.

Steps to reproduce:

1. Run Optimize with multiple definitions in any mode
2. Observe amount of requests to proces overview index - it should be one per definition every 10 mins (or more frequently if KPI interval is configured differently)

Actual result:

We observe way more requests to the process overview index than expected. 

Expected result:

We observe one request per process per KPI evaluation round.

Slack: #support-16769-operate-responds-with-500-http-code

Note: this needs backporting – check when the issue was introduced and backport accordingly.